Pharmaceutical Research, Vol. 8, No. 3, 1991
`
`Review
`
`Protein—Solvent Interactions in Pharmaceutical Formulations
`
`Tsutomu Arakawa,1’3 Yoshiko Kita,1 and John F. Carpenter2
`
`The stability of proteins is affected by a variety of solvent additives. Sugars, certain amino acids and
`salts, and polyhydric alcohols stabilize proteins in solution and during freeze-thawing. Urea and
`guanidine hydrochloride destabilize proteins under either condition. These effects can be explained
`from the preferential interactions of the cosolvents with the proteins; i.e., the protein stabilizers are
`preferentially excluded from the proteins, while the destabilizers bind to them. There is a class of
`compounds, such as polyethylene glycol and 2-methyl-2,4-pentanediol, that destabilize proteins at
`high temperature but stabilize them during freeze-thawing. Such effects can be accounted for by their
`preferential exclusion from the native proteins determined at room temperature and from their hy-
`drophobic character, which depends on temperature. During freeze-drying, only a few sugars appear
`to be effective in protecting proteins from inactivation, as most other stabilizers cannot exert their
`action on proteins without water. The stabilization is due to hydrogen bonding between the sugars and
`the dried proteins, the sugars acting as water substitute. Understanding the mechanism of the effects
`of solvent additives on the protein stability should aid in the development of a suitable formulation for
`proteln.
`
`KEY WORDS: protein—solvent interaction; protein stability; freeze-thawing; freeze-drying; hydro-
`phobic interaction.
`
`INTRODUCTION
`
`Proteins are marginally stable under physiological con—
`ditions, the free energy of denaturation being 5 to 20 kcal/
`mol (1—4). This is because stabilizing free energy, arising
`from interresidue interactions in a globular structure,
`is
`largely canceled by a loss of entropy that arises from the
`protein’s compactness. A variety of compounds has been
`shown to affect the stability of proteins in solution. Sugars,
`polyols, and certain amino acids and salts are known to be
`protein stabilizers (5—12). On the other hand, hydrophobic
`organic compounds, chaotropic salts, urea, and guanidine
`hydrochloride are known to be protein destabilizers (6,7,13—
`15). The above protein stabilizers are also effective in the
`frozen state as well as in solution. Certain organic com—
`pounds such as dimethyl sulfoxide, 2—methyl-2,4-pentane-
`diol, and polyethylene glycol are also effective as protein
`stabilizers in the frozen state, but not in solution (in partic-
`ular at higher temperatures), whereas urea and guanidine
`hydrochloride are destabilizers in the frozen state as well.
`Although a broad range of compounds can serve as pro-
`tein stabilizers in solution and frozen state, Carpenter and
`Crowe (16) have recently shown that only carbohydrates can
`protect a labile enzyme (i.e, phosphofructokinase) from in-
`activation by freeze—drying. This suggests that a fundamental
`difference exists in stabilizing mechanism between the dried
`
`1 Amgen Inc., Amgen Center, Thousand Oaks, California 91320.
`2 CryoLife, Inc., 2211 New Market Parkway, Suite 142, Marietta,
`Georgia 30067.
`3 To whom correspondence should be addressed.
`
`state and the solution or frozen state, as has recently been
`discussed by Crowe et al. (17).
`We summarize the solvent additives that have been
`used to enhance the stability of proteins in solution and dur-
`ing freeze-thawing and freeze-drying and then describe the
`current knowledge of the mechanism of their actions. Fi—
`nally, we attempt to show how one can use this knowledge to
`achieve suitable formulations for protein pharmaceuticals.
`
`Cosolvent Effects in Solution
`
`Examples of cosolvent—induced stabilization of proteins
`in solution may be arbitrarily divided into two groups, i.e.,
`stabilization against reversible denaturation induced by dif-
`ferent types of perturbation and stabilization against time-
`dependent irreversible denaturation or slow reversible dena-
`turation. In their classical work, von Hippel and Wong (12)
`showed that potassium phosphate and ammonium sulfate in-
`crease the transition temperature of the globular protein,
`ribonuclease A, and hence stabilize it from a reversible de—
`naturation, NaCl and KCl have no effect, and CaCl2 and
`KSCN are destabilizers. Similar effects of the salts were
`
`observed on the stability of a nonglobular protein, collagen
`(11), and on the thermal stability of myosin (18). In the latter
`study, the bromide salts were the strongest destabilizers;
`sodium citrate and the other stabilizing salts increased the
`transition temperature. Salt-induced stabilization and desta-
`bilization can be found with many other proteins (19—21).
`B-Lactoglobulin can be stabilized against urea-induced un-
`folding by salts (at 0.2 M), the stabilization following the
`order of chloride < tartrate < sulfate < phosphate < citrate
`
`285
`
`0724-874l/9l/0300-0285$06.50/0 © 1991 P1
`
`AMGEN INC.
`
`Exhibit 1025
`
`Ex. 1025 - Page 1 of8
`
`Ex. 1025 - Page 1 of 8
`
`AMGEN INC.
`Exhibit 1025
`
`

`

`Pharmaceutical Research, Vol. 8, No. 3, 1991
`
`Review
`
`Protein—Solvent Interactions in Pharmaceutical Formulations
`
`Tsutomu Arakawa,1’3 Yoshiko Kita,1 and John F. Carpenter2
`
`The stability of proteins is affected by a variety of solvent additives. Sugars, certain amino acids and
`salts, and polyhydric alcohols stabilize proteins in solution and during freeze-thawing. Urea and
`guanidine hydrochloride destabilize proteins under either condition. These effects can be explained
`from the preferential interactions of the cosolvents with the proteins; i.e., the protein stabilizers are
`preferentially excluded from the proteins, while the destabilizers bind to them. There is a class of
`compounds, such as polyethylene glycol and 2-methyl-2,4-pentanediol, that destabilize proteins at
`high temperature but stabilize them during freeze-thawing. Such effects can be accounted for by their
`preferential exclusion from the native proteins determined at room temperature and from their hy-
`drophobic character, which depends on temperature. During freeze-drying, only a few sugars appear
`to be effective in protecting proteins from inactivation, as most other stabilizers cannot exert their
`action on proteins without water. The stabilization is due to hydrogen bonding between the sugars and
`the dried proteins, the sugars acting as water substitute. Understanding the mechanism of the effects
`of solvent additives on the protein stability should aid in the development of a suitable formulation for
`proteln.
`
`KEY WORDS: protein—solvent interaction; protein stability; freeze-thawing; freeze-drying; hydro-
`phobic interaction.
`
`INTRODUCTION
`
`Proteins are marginally stable under physiological con—
`ditions, the free energy of denaturation being 5 to 20 kcal/
`mol (1—4). This is because stabilizing free energy, arising
`from interresidue interactions in a globular structure,
`is
`largely canceled by a loss of entropy that arises from the
`protein’s compactness. A variety of compounds has been
`shown to affect the stability of proteins in solution. Sugars,
`polyols, and certain amino acids and salts are known to be
`protein stabilizers (5—12). On the other hand, hydrophobic
`organic compounds, chaotropic salts, urea, and guanidine
`hydrochloride are known to be protein destabilizers (6,7,13—
`15). The above protein stabilizers are also effective in the
`frozen state as well as in solution. Certain organic com—
`pounds such as dimethyl sulfoxide, 2—methyl-2,4-pentane-
`diol, and polyethylene glycol are also effective as protein
`stabilizers in the frozen state, but not in solution (in partic-
`ular at higher temperatures), whereas urea and guanidine
`hydrochloride are destabilizers in the frozen state as well.
`Although a broad range of compounds can serve as pro-
`tein stabilizers in solution and frozen state, Carpenter and
`Crowe (16) have recently shown that only carbohydrates can
`protect a labile enzyme (i.e, phosphofructokinase) from in-
`activation by freeze—drying. This suggests that a fundamental
`difference exists in stabilizing mechanism between the dried
`
`1 Amgen Inc., Amgen Center, Thousand Oaks, California 91320.
`2 CryoLife, Inc., 2211 New Market Parkway, Suite 142, Marietta,
`Georgia 30067.
`3 To whom correspondence should be addressed.
`
`state and the solution or frozen state, as has recently been
`discussed by Crowe et al. (17).
`We summarize the solvent additives that have been
`used to enhance the stability of proteins in solution and dur-
`ing freeze-thawing and freeze-drying and then describe the
`current knowledge of the mechanism of their actions. Fi—
`nally, we attempt to show how one can use this knowledge to
`achieve suitable formulations for protein pharmaceuticals.
`
`Cosolvent Effects in Solution
`
`Examples of cosolvent—induced stabilization of proteins
`in solution may be arbitrarily divided into two groups, i.e.,
`stabilization against reversible denaturation induced by dif-
`ferent types of perturbation and stabilization against time-
`dependent irreversible denaturation or slow reversible dena-
`turation. In their classical work, von Hippel and Wong (12)
`showed that potassium phosphate and ammonium sulfate in-
`crease the transition temperature of the globular protein,
`ribonuclease A, and hence stabilize it from a reversible de—
`naturation, NaCl and KCl have no effect, and CaCl2 and
`KSCN are destabilizers. Similar effects of the salts were
`
`observed on the stability of a nonglobular protein, collagen
`(11), and on the thermal stability of myosin (18). In the latter
`study, the bromide salts were the strongest destabilizers;
`sodium citrate and the other stabilizing salts increased the
`transition temperature. Salt-induced stabilization and desta-
`bilization can be found with many other proteins (19—21).
`B-Lactoglobulin can be stabilized against urea-induced un-
`folding by salts (at 0.2 M), the stabilization following the
`order of chloride < tartrate < sulfate < phosphate < citrate
`
`285
`
`0724-874l/9l/0300-0285$06.50/0 © 1991 Plenum Publishing Corporation
`
`Ex. 1025 - Page 2 of 8
`
`Ex. 1025 - Page 2 of 8
`
`

`

`286
`
`(19). Antithrombin III, which undergoes heat-induced dena-
`turation followed by the formation of aggregates, is strongly
`stabilized by phosphate and sulfate and destabilized by io-
`dide and thiocyanate. It is noteworthy that ethylenedi-
`aminetetraacetic acid and citrate are extremely strong stabi-
`lizers.
`
`Effects of polyhydric alcohols on the thermal stability of
`globular proteins were extensively studied by Gerlsma (6,7).
`Glycerol, erythritol, and sorbitol increase the transition tem-
`perature of the protein regardless of pH, the increase being
`greater at higher cosolvent concentrations. Stabilizing ef-
`fects of polyhydric alcohols have been demonstrated with
`collagen (23,24), chymotrypsinogen (7), lysozyme (22) and
`other proteins (25). Ethylene glycol decreases the transition
`temperature of ovalbumin, while the effect on ribonuclease
`A is more complex, being a protein stabilizer at pH 2.3 but a
`destabilizer at pH 5.5. This suggests that the stabilizing ef-
`fect of ethylene glycol is variable; it induces concentration—
`dependent denaturation of B—lactoglobulin (14) and de-
`creases the thermal stability of lysozyme (22), while it in—
`creases the transition temperature of collagen (23).
`Sugars are also effective protein stabilizers as measured
`by their ability to increase the transition temperature of col-
`lagen (23). Sucrose and maltose are most effective, ribose
`and deoxyribose least effective, and glucose and mannose
`are intermediate. The transition temperature of ovalbumin
`increases in the presence of various sugars, including arab-
`inose, glucose, galactose, and mannose. A polysaccharide,
`dextran 10, but not Ficoll, also stabilizes the protein (15).
`Lee and Timasheff (26) found that sucrose increases the
`transition temperatures of chymotrypsinogen, a-chymo-
`trypsin, and ribonuclease. These results demonstrate that
`the effects of sugars are general and independent of the pro-
`teins used.
`
`Sugars also have been shown to protect proteins from
`time—dependent denaturation. The effect of varying concen-
`trations of sucrose on the colchicine binding ability of tubu—
`lin has been studied (27). Tubulin can bind colchicine only
`when it is in the native state. Sucrose (1 M) can give com-
`plete protection from inactivation, while at lower sucrose
`concentrations there is a gradual loss of the ability of tubulin
`to bind colchicine. Glucose also can stabilize tubulin against
`time-dependent denaturation (28).
`Certain amino acids and amine compounds also serve as
`protein-stabilizing cosolvents. The transition temperatures
`of lysozyme were determined in the absence and presence of
`various amino acids (e. g., L—proline, L-serine, or-alanine, and
`B-alanine) and related compounds (e.g., trimethylamine N—
`oxide, taurine, 'y-aminobutyric acid, sarcosine, and betaine)
`(8). The transition temperature was increased by the inclu-
`sion of these additives. Monosodium glutamate, lysine hy-
`drochloride, glycine, and betaine at
`1 M also have been
`shown to increase the thermal stability of bovine serum al—
`bumin and lysozyme (9,10).
`These compounds, at moderate concentrations, also
`protect proteins from time-dependent denaturation and in—
`activation and offset the deleterious effects of urea on pro-
`tein stability (29—31). Monosodium glutamate (28) and
`e-aminocaproic acid (32), respectively, can protect tubulin
`and reduce degradation of allergen component proteins in
`aqueous pollen extracts.
`
`Arakawa, Kita, and Carpenter
`
`All of the compounds described above, except ethylene
`glycol, also enhance self-association of proteins, protein—
`protein interactions, and interactions of proteins with other
`ligands and decrease protein solubility. Sucrose, glycerol,
`and monosodium glutamate enhance microtubule assembly
`(56,59—61), and structure-stabilizing salts enhance actin
`polymeration (62), B-lactoglobulin dimerization (19), and he-
`mocyanin self-association (63). Polyethylene glycol is widely
`used as a protein precipitant and to enhance protein—protein
`interactions, yet it is not a protein stabilizer; it decreases the
`transition temperature of some monomeric proteins (64,65).
`Polyethylene glycol 8000 (PEG 8000) enhances the associa—
`tion between glycolytic enzymes and between glycolytic en-
`zymes and F—actin (66), among other proteins (67,68).
`
`Cosolvent Effects During Freeze-Thawing
`
`The bulk of the research on protein cryopreservation
`has been on enzymes, where stabilization is reflected in the
`maintenance of catalytic activity upon thawing. Protection
`against freeze-thawing can be afforded by the compounds
`described above and even other proteins (e.g., bovine serum
`albumin) (33—35). Usually cosolvent concentrations exceed-
`ing 0.2 M are needed for cryopreservation.
`Loomis et al. (36,37) have demonstrated that end prod—
`ucts of anaerobic metabolism, which are thought to protect
`certain organisms against freeze-induced damage in nature,
`can also stabilize labile enzymes such as lactate dehydroge-
`nase and phosphofructokinase. These compounds include
`strombine, alanopine, octopine, lactate, succinate, and pro—
`pionate.
`Carpenter and colleagues (38—41) have shown that com-
`binations of certain divalent cations (e. g. , an +) and organic
`cosolvents (e.g., sugars, polyhydric alcohols, amino acids,
`and related compounds) can provide synergistic protection
`for labile enzymes and antibodies. For example, full activity
`of phosphofructokinase is recovered when this sensitive en-
`zyme is frozen in the presence of 0.6 mM zinc sulfate and 5
`mM trehalose. With either additive alone, no enzyme activ—
`ity is measurable after freeze-thawing. This phenomenon
`could have practical advantages in protein formulation since
`reduced amounts of stabilizers could be used.
`
`Cosolvent Effects During Freeze— and Air-Drying
`
`Many cryoprotectants are unable to stabilize labile en-
`zymes during drying and rehydration (17,35). Carpenter and
`colleagues found that only carbohydrates stabilize phospho-
`fructokinase during either freeze-drying or air-drying. The
`greatest degree of protection is seen with the disaccharides,
`trehalose, sucrose, maltose, and lactose (16,42—44).
`Monosaccharides (e.g., glucose and galactose) and myo-
`inositol are much less effective.
`
`Synergistic stabilization of phosphofructokinase can be
`realized during freeze-drying and air-drying with combina-
`tions of divalent zinc and carbohydrates (42,43). However,
`even in the presence of zinc, cryoprotectants such as amino
`acids and glycerol do not preserve the enzyme in the dried
`state.
`
`MECHANISM OF STABILIZATION OF PROTEINS
`BY COSOLVENTS
`
`Since a broad spectrum of cosolvents can stabilize pro—
`
`Ex. 1025 - Page 3 of78,
`
`Ex. 1025 - Page 3 of 8
`
`

`

`Protein—Solvent Interactions
`
`287
`
`Table I. Preferential Interactions of Cosolvents with Proteins
`
`“ S
`
`teins under widely different conditions, it is unlikely that the
`stabilization of proteins by the cosolvents in solution and
`during freeze-thawing stems from a specific binding effect.
`The preferential interaction of proteins with solvent compo-
`nents can explain, probably without exception, the effects of
`cosolvents on the stability of proteins both in solution and
`during freeze-thawing (33,45).
`Here we describe the preferential interaction of proteins
`with solvent components and the thermodynamic conse-
`quences. The protein—ligand interaction is described by the
`bindings of water (designated with the subscripts w) and
`ligand (s) to a protein (p), according to Eq. (1); at constant
`temperature, pressure, and chemical equilibrium;
`
`(figs/6gp)T,uw,us : gwAs “ gsAw
`
`(1)
`
`where g, is the concentration of component i (s, p, or w) in
`grams per gram of water in the system, T is Kelvin temper—
`ature, u, is the chemical potential of component i, and As and
`AW are the bindings of ligand and water expressed as gram
`per gram of protein. At high ligand concentrations, the sec—
`ond term, i.e. , the binding of water, contributes significantly
`to the observed value. Under such conditions, the preferen-
`tial binding, (Egg/6gp) (subscripts omitted), becomes different
`from the total ligand binding, As. The parameter (6gs/6gp) can
`be either positive or negative. When the parameter is nega-
`tive, water is in excess in the protein domain over its con-
`centration in the bulk, i.e., the protein is preferentially hy-
`drated. The preferential hydration parameter (agw/agp) can
`be calculated from
`
`(agw/agp) = “ (l/gs)(ags/agp)
`
`(2)
`
`The preferential interaction parameter expressed in grams
`can be converted to that in moles by Eq. (3):
`
`(ems/amp) = (Mp/Msxags/agp)
`
`(3)
`
`where M,- is the molecular weight, and m, is the molal con-
`centration of component i.
`Preferential interaction parameters have been deter-
`mined for salts (46—54), amino acids and related compounds
`(8,52), polyhydric alcohols (55—57), and sugars (26,58) with
`various proteins. Typical results for salts, amino acids, and
`sugars are shown in Tables I and II. These compounds have
`large negative values of (figs/6gp). The structure stabilizing
`compounds are excluded from the protein surface. The
`structure destabilizing compounds, such as MgCl2 and
`KSCN, have been shown to be bound to the proteins
`(49,50,53).
`Table III lists the preferential interaction values of sev-
`eral polyhydric alcohols with bovine serum albumin; the pro-
`tein is preferentially hydrated and these cosolvents are ex-
`cluded from the protein. The preferential exclusion of glyc—
`erol was also observed with tubulin and other proteins
`(56,57). Ethylene glycol shows a small preferential exclusion
`from bovine serum albumin, yet binds to B-lactoglobulin
`(14), indicating that the preferential interaction of ethylene
`glycol may depend on the kind of protein. This observation
`is consistent with the fact that ethylene glycol stabilizes
`some proteins while destabilizing others.
`Preferential interaction simply reflects the perturbation
`
`(egg/6gp)
`Protein
`olvent
`~ 0.106
`Tubulin
`1 M sucrose
`Ribonuclease A —0.190
`1 M sucrose
`BSA"
`— 0.099
`2 M glucose
`BSA
`—0.113
`3.4 Mglycerol
`BSA
`— 0.069
`2 M glycine
`BSA
`— 0.125
`2 M betaine
`2 M Na glutamate BSA
`—0.l7l
`1 M Na glutamate BSA
`—0.088
`1 M Na glutamate Tubulin
`~0.058
`1 M NaCl
`BSA
`—0.0145
`1 M NaZSO4
`BSA
`—0.074
`1 M NaOAc
`BSA
`—0.027
`1 M MgSO4
`BSA
`—0.047
`
`(ems/amp)
`— 38.0
`—7.6
`— 37.4
`~83.4
`— 62.2
`— 72.5
`—68.8
`—35.5
`— 37.3
`— 16.8
`—35.4
`— 22.4
`— 26.5
`
`(agw/agp)
`0.243
`0.437
`0.212
`0.212
`0.416
`0.428
`0.417
`0.477
`0.393
`0.243
`0.524
`0.312
`0.388
`
`‘1 Bovine serum albumin.
`
`of the chemical potential of the protein by the ligand (69,70).
`Equation (4) shows that a negative value of (ems/amp)
`
`(ams/amp) = _ (aHp/amsywl-Ls/ams)
`
`(4)
`
`means that (app/ems) is positive; namely, the addition of
`ligand (stabilizer) increases the chemical potential of the pro—
`tein and, thus, the free energy of the system. This is ther-
`modynamically unfavorable. If, in the course of the denatur-
`ation reaction, the chemical nature of the interactions be-
`tween the protein and the stabilizer does not change, this
`situation should become even more thermodynamically un—
`favorable for the unfolded state of the protein due to the
`increase in the surface area of contact between protein and
`solvent. Therefore, the reaction is pushed toward the native
`state, resulting in stabilization of the native structure.
`For a wide variety of compounds, exclusion is deter-
`mined by the effect of the additive on the surface tension of
`water. Cosolvents perturb the cohesive force of water and,
`hence, its surface tension. This results in either an excess or
`a deficiency of the cosolvent in the protein surface layer (71).
`Those compounds which increase the surface tension of wa-
`ter should be excluded from the protein surface (49—53).
`Although exclusion predominates for the structure sta-
`bilizing compounds, nevertheless these can bind to proteins
`through hydrophobic interaction, hydrogen bonding, or elec—
`trostatic interactions. The net interaction observed between
`proteins and stabilizing compounds is the balance between
`binding to the protein and exclusion. If the binding of a co-
`solvent increases more than the exclusion does upon dena-
`turation, the result should be protein destabilization, while
`
`Table II. Interaction Parameters of Lysozyme with Amino Acids
`and Related Compounds at pH 6.0W
`
`
`
` Solvent (6g5/6gp) (g/g) (agw/ng) (g/g)
`
`
`
`0.322 t 0.074
`—0.0406 i 0.0093
`1 M L-proline
`0.444 t 0.027
`—0.0497 x 0.0030
`1 M L-serine
`0.377 t 0.058
`—0.0331 : 0.0051
`0.667 M taurine
`0.629 t 0.074
`-—0.0699 : 0.0082
`1 M y-aminobutyric acid
`0.485 1 0.040
`—0.046l : 0.0038
`1 M sarcosine
`m
`
`Ex. 1025 - Page 4 of 8
`
`Ex. 1025 - Page 4 of 8
`
`

`

`288
`
`Table III. Preferential Interaction Parameters of Bovine Serum Al-
`bumin with Solvent Components in Water—Polyhydric Alcohol Sys-
`tems at 25°C
`
`
`(egg/6gp)
`g/g
`
`(agw/agp)
`g/g
`
`(amw/amp)
`mol/mol
`
`Water—ethylene glycol
`
`Alcohol
`(%)
`
`0
`20 (v/v)
`40
`60
`
`10 (v/V)
`20
`30
`40
`
`—0.041 i 0.020
`—0.097 t 0.027
`—0.222 r 0.041
`
`Water—glycerol
`
`—0.020 t 0.010
`~0.052 : 0.023
`—0.101 t 0.020
`—0.154 t 0.024
`
`Water—xylitol
`
`20 (w/v)
`30
`
`—0.030 t 0.021
`—0.055 t 0.011
`
`Water—mannitol
`
`10 (w/v)
`15
`
`70.022 : 0.006
`—0.034 t 0.007
`
`Water—sorbitol
`
`5 (w/v)
`10
`15
`20
`30
`40
`
`~0.009 : 0.009
`—0.022 : 0.009
`—0.033 t 0.010
`—0.054 i 0.011
`—0.092 t 0.016
`—0.128 : 0.018
`
`Water—inositol
`
`0.148
`0.130
`0.137
`
`0.143
`0.165
`0.187
`0.185
`
`0.129
`0.146
`
`0.205
`0.204
`
`0.174
`0.205
`0.198
`0.229
`0.245
`0.234
`
`560
`490
`520
`
`540
`620
`710
`700
`
`490
`550
`
`770
`770
`
`660
`770
`750
`870
`930
`880
`
`1,540
`0.407
`—0.021 t 0.003
`5 (w/v)
`
`
`
`—0.041 r 0.005 0.38710 1,460
`
`the opposite should be true for the structure stabilizers. For
`strong protein denaturants, binding always predominates.
`The effect of stabilizers on a reaction involving protein—
`protein contacts, such as protein self-assembly or precipita-
`tion, may be viewed in the same way as described for the
`denaturation reaction. Exclusion of the stabilizer per mono—
`meric protein unit is decreased upon formation of protein—
`protein contacts; namely, the associated form of the precip-
`itate is less unfavorable thermodynamically in the presence
`of the stabilizer.
`
`The mechanism of protein stabilization described above
`should apply both in solution and during freeze—thawing.
`However, polyethylene glycol and 2-methyl-2,4-pentanediol
`(MPD) are unique in that they are preferentially excluded
`from the protein, but they denature or destabilize proteins in
`solution. Nevertheless, they can stabilize proteins during
`freeze-thawing.
`The postulated mechanism by which these compounds
`exert such a unique effect (34,54) is through exclusion from
`the native protein due to steric exclusion and/or repulsion
`from protein charges and binding to the denatured protein
`due to hydrophobic interaction. Therefore, the cosolvent ex-
`clusion will increase upon denaturation, but the cosolvent
`
`Arakawa, Kita, and Carpenter
`
`binding will do so even more strongly, and the net result is a
`decrease in the preferential exclusion upon denaturation.
`How do these cosolvents stabilize protein during freeze-
`thawing? At subzero temperatures, the hydrophobic charac-
`ter of these compounds will become weak, and hence hy-
`drophobic interaction between the proteins and these com-
`pounds, which is the driving force of destabilization, will no
`longer contribute to the preferential interaction. Conse-
`quently, preferential exclusion of these compounds will
`dominate for the native and denatured states of the protein,
`and hence the net result is stabilization. The same argument
`should apply to the proteins undergoing self-association,
`since these compounds stabilize monomeric proteins and en-
`hance their associations at subzero temperatures.
`We now turn to the mechanism of cosolvent effect on
`
`proteins during freeze—drying. As noted above, only carbo-
`hydrates are effective at protecting phosphofructokinase
`during either freeze-drying or air-drying (42,43). The obser-
`vation that many effective cryoprotectants fail to protect
`dried phosphofructokinase indicates that the mechanism of
`cosolvent-induced protein stabilization in the dried state is
`fundamentally different from that for proteins in aqueous or
`frozen systems (16,17,44). In addition, the thermodynamic
`arguments that are needed to explain protein stabilization by
`preferentially excluded cosolvents are not applicable when
`water is removed from the system.
`It has been suggested (44) that certain carbohydrates
`protect proteins by binding to the dried protein, thus serving
`as a “water substitute,” when the hydration shell of the
`protein is removed. Fourier transform infrared spectroscopy
`(16) indicated not only that hydrogen bonding occurs be-
`tween proteins and stabilizing carbohydrates but also that
`solute binding is requisite for labile proteins to be preserved
`during drying. For example, the presence of the protein (bo-
`vine serum albumin or lysozyme) leads to a pronounced de—
`crease in absorbance in the fingerprint region of the infrared
`spectrum for trehalose, major shifts in band position, and a
`loss of band splitting. The protein-induced spectral changes
`can be titrated by freeze-drying the sugar with increasing
`amounts of either protein (18).
`The significance of the influence of proteins on the vi-
`brational spectrum of dried trehalose can best be appreciated
`when compared to the effects of water on the spectrum of
`the hydrated sugar. Typical spectra for trehalose dried in the
`presence of either lysozyme or bovine serum albumin are
`remarkably similar to that for hydrated trehalose, while all
`are very different from a spectrum of crystalline trehalose.
`Thus, it appears that proteins serve the same role for dried
`trehalose as does water for the hydrated sugar, i.e., they
`form hydrogen bonds with the polar groups in the sugar.
`It is implicit in the conclusion that proteins serve as
`water substitutes for dried carbohydrates that the converse
`must also be true. To test this suggestion, the influence of
`trehalose on the infrared spectrum of lysozyme was investi-
`gated (16). When lysozyme is dried without the sugar, there
`is an increase in the frequency of the amide I band from
`1652.1 cm”, seen for the fully hydrated protein, to 1659
`cm' 1. The amide II band is broadened and shifts from about
`1543 to almost 1530 cmT1 in the dried protein. In addition,
`the band assigned to the carboxylate in the hydrated protein
`(72) at 1583 cm‘1 is not detectable with the dried protein.
`
`EX. 1025 - Page 5 91°37",
`
`Ex. 1025 - Page 5 of 8
`
`

`

`Protein—Solvent Interactions
`
`289
`
`trehalose during sublimation could decrease the availability
`of the sugar for forming hydrogen bonds with the protein
`(16).
`
`Significantly, the greatest degree of protection of phos-
`phofmctinase during freeze drying (16) is noted with inter—
`mediate amounts of sugar (Fig. 13). With trehalose concen-
`trations greater than 150 mg/ml, there is a decrease in activ-
`ity recovered. These results indicate that whenever
`trehalose is at a concentration that does not influence the
`
`frequency of the amide II band for dried lysozyme, this sugar
`concentration is also ineffective at preserving dried, labile
`proteins. That is, hydrogen bonding of the sugar to the pro-
`tein appears to be mandatory for the sugar to preserve dried
`proteins.
`During freeze-thawing, however, the presence of high
`concentrations of trehalose leads to increased recovery of
`activity (Fig. 1B), as expected, since cryopreservation is due
`to the preferential exclusion of the stabilizing cosolvent from
`the surface of the protein.
`In summary, sugars such as trehalose can serve to sat-
`isfy partially the hydrogen-bonding requirements of the polar
`groups on dried proteins and, thus, serve as water substi-
`tutes. The carbohydrate may prevent the formation of intra-
`and interprotein hydrogen bonding in the dried state, which
`could induce unfolding and/or aggregation of protein mole-
`cules upon rehydration. It is clear that this is distinctly dif—
`ferent from the preferential exclusion mechanism that is op-
`erative in the aqueous and frozen state. Therefore,
`it is
`important, for discussions regarding stabilization of biomol-
`ecules, that freeze-thawing and freeze-drying, although both
`dependent on a freezing step, be viewed as distinctly differ-
`ent stress vectors (cf. Refs. 16, 17, 35).
`
`PRACTICAL CONSIDERATIONS
`
`It is clear that certain salts, sugars, and amino acids and
`related compounds can protect proteins from denaturation
`or inactivation in solution and during freeze-thawing. Car-
`bohydrates can also protect proteins from damages that oc-
`cur during drying. Since these compounds stabilize proteins
`at high and low temperatures, it is not absolutely necessary
`to keep the protein solution at low temperatures. However,
`those compounds which destabilize the proteins at high tem-
`peratures (e.g., PEG and MPD) but stabilize them at low
`temperature must be used with caution.
`All the compounds described above enhance protein—
`protein and protein—ligand interactions. Therefore, they
`might enhance the binding of proteins to glass vials or lead to
`a decrease in the protein solubility. Polyethylene glycol is
`often used to prevent proteins from binding to glass during
`sample storage and column chromatography. The denatur-
`ation action of polyethylene glycol at high temperatures is
`extremely weak (37,48,49), but its precipitating action is ex-
`tremely strong. The reason why this compound does not
`enhance the protein binding to the glass is probably due to
`the binding affinity of polyethylene glycol itself to the glass
`surface. Therefore, although polyethylene glycol is not a
`protein stabilizer in solution (in particular at high tempera-
`ture), it may be used to reduce surface adsorption of proteins
`and stabilize proteins from freeze-thawing, provided that ap-
`
`Ex. 1025 - Page 6 of 8
`
`When lysozyme is freeze-dried in the presence of trehalose
`(5 g sugar/g protein), the amide I band of the dried protein is
`shifted back to 1658.1 cm‘ 1. A band corresponding to that
`for carboxylate at 1583 cm’1 also appears. The most dra—
`matic effect of the sugar is to shift the amide II band back to
`1542 cm”, almost the identical position noted for the hy—
`drated protein. In addition, the band shape is essentially the
`same as that seen for the hydrated protein.
`The position of the amide II band was used as a means
`to characterize the effect of varying the amount of sugar on
`the vibrational spectrum of dried lysozyme. Figure 1A
`shows there is an increase in the amide II frequency as the
`initial sugar concentration is increased up to 100 mg/ml. With
`100—200 mg/ml trehalose, the amide II band is centered at
`about 1542 cm". However, when the sugar concentration is
`greater than 200 mg/ml, there is a p